LITTORALICIOUSIn This Article
Why This Matters for Yacht Chefs
You already work 16-hour days. You already juggle guest preferences, dietary restrictions, provisioning logistics, crew meals, and themed dinner parties — often simultaneously, often alone, often in a galley the size of a bathroom.
AI does not replace your palate, your instinct, or your ability to read a room. What it does is handle the parts of your job that are administrative, repetitive, and time-consuming — so you can spend more time on the parts that actually require a chef.
The principle: AI is a tool. Like a Thermomix or a Pacojet. It does what you tell it. It gets better the more precisely you instruct it. And like every tool in your galley, it’s only as good as the person using it.
What AI Can Do in Your Galley
Menu creation. Tell it the cuisine, the proteins you have, the guest preferences — and it builds a complete menu card with Cultural Touchstone naming, ingredient-first descriptions, and production-ready PDF output. In minutes, not hours.
Guest preference management. Paste in WhatsApp messages, trip feedback, or preference sheets. The AI extracts every like, dislike, allergy, and dietary requirement — then cross-references them against every menu you create.
Recipe development. “Give me a branzino dish that works with the Italian theme, avoids onions, and uses what’s in season in the Mediterranean in March.” Done.
Provisioning. “I’m arriving in Palma with 8 guests for 5 days. Build me a provision list for Mediterranean and Japanese themes.” It knows the maths.
Wine pairing. One wine per menu, from sustainable producers with real stories — not the obvious bottle, but the one that makes guests ask where you found it.
Translation and cultural research. Hosting Greek guests? Ask for traditional dishes, proper pronunciation, and regional variations. The AI knows more cuisines than any single chef ever will.
Real Menus, Real Prompts
This is not theoretical. Here are real prompts from a working galley — and the kind of menus that came back in under two minutes.
Crew Lunch: Make-Your-Own Deli
The AI builds the full spread: bread options, proteins, cheeses, condiment bar, two salads, a soup. You adjust, confirm, done. Your crew gets variety. You spend 20 minutes instead of 40.
Crew Lunch: Poke Bowl Bar
Back comes a complete station: base options (sushi rice, brown rice, mixed greens), proteins, 8–10 toppings, three sauces with recipes. The kind of crew lunch that makes people actually say thank you.
Guest Dinner: Nobu Style
The AI returns a 7-course tasting menu with a Cultural Touchstone name, each dish described ingredient-first, a gluten-free footnote only where needed, and a sake pairing suggestion. Print-ready in minutes.
More Things You Can Ask
- “Build me a Mediterranean family-style lunch. I have lamb, branzino, and lots of summer vegetables.”
- “Crew breakfast for 7, make-your-own açaí bowl station. What toppings should I prep?”
- “I’m doing a BBQ for 12 guests tonight. American steakhouse theme. What cuts, what sides, what order?”
- “Convert last night’s Italian dinner into a provision list for 5 days.”
- “Give me 5 Cultural Touchstone names for a Greek island sunset dinner.”
- “The guests loved the tuna tartare. Give me 3 variations I can rotate through the week.”
The pattern: Be specific about what you have, who you’re feeding, and the style you want. The more context you give, the less you need to correct. Talk to it like a commis on day one — clear, direct, no ambiguity.
What the Menus Actually Look Like
These are real menus generated by AI and formatted as print-ready PDF cards using LaTeX typesetting. Each one took under two minutes from prompt to print. You describe what you want — the system handles fonts, layout, and formatting.
San Marzano tomato sauce, fresh basil, aged Parmigiano-Reggiano, extra-virgin olive oil
hand-breaded, mozzarella di bufala, slow-simmered marinara, fresh oregano
garlic, chili flakes, lemon zest
anchovy, garlic, olive oil
sourdough, roasted garlic butter, parsley
burrata, heirloom tomato, fresh basil, aged balsamic, extra-virgin olive oil
Each of these menus was generated in under 2 minutes. The Cultural Touchstone name, ingredient descriptions, and layout — all from a single prompt. For PDF output with professional typography, see Claude Code in the Galley.
How to Get Started (3 Minutes)
You have two paths. Both work. Choose the one that fits how you think.
Path A: The Browser (Easiest)
No installation. No terminal. Just a browser.
1 Go to claude.ai (or chatgpt.com, or gemini.google.com)
2 Create a free account (or subscribe for more power)
3 Copy the Menu Agent prompt from below
4 Paste it into the chat as your first message
5 Then talk to it: “Guests dinner, Italian, I have branzino and lamb”
That’s it. The AI now has your entire menu system loaded and will behave exactly like the instructions say — Cultural Touchstone names, ingredient-first, no operational noise on the card.
Path B: Claude Code (More Powerful)
If you want the AI to actually create files, generate PDFs, read your guest preference sheets, and manage your entire galley document system — Claude Code is the tool. It runs in your terminal and works directly with your files. See Claude Code in the Galley for the full setup guide.
The Menu Agent: Copy, Paste, Cook
This is the complete menu system we use. It works in any AI platform — Claude, ChatGPT, Gemini, Copilot, Mistral, or any other LLM. Copy the entire prompt below, paste it as your first message in a new conversation, and the AI becomes your menu architect.
How to use it: Copy the prompt below. Open any AI chat. Paste the entire thing as your first message. Then tell it what you want: “Guests lunch, Thai theme, I have salmon and prawns.” The AI handles the rest.
Why paste the whole prompt? AI systems have no memory between conversations. Each new chat starts blank. Pasting the prompt first gives the AI its “training” for that session — like giving a new commis your standards on day one. The more specific your instructions, the better the output. This prompt has been refined over 50+ menu sessions.
Where to Use It
The menu agent works on any AI platform. Copy the prompt, paste it, and start talking. Here are the main options:
By Anthropic. Best for long, detailed instructions. Follows complex prompts precisely. Our recommendation.
By OpenAI. Most widely used. Good general performance. Free tier available.
By Google. Integrated with Google Workspace. Good for research-heavy tasks.
By Microsoft. Built into Windows and Office. Convenient if you already use Microsoft tools.
Pro tip: On Claude, use “Projects” to save your menu agent as a persistent system prompt. Every new conversation in that project will already have the agent loaded — no re-pasting needed.
Privacy and Your Guests’ Data
This matters. You handle sensitive information — guest names, dietary restrictions, allergies, personal preferences, travel plans. Before you paste any of that into an AI, you need to know where it goes.
Rule of thumb: Never put a guest’s full name, contact details, or anything that identifies them personally into a free-tier AI chat. If you need to reference preferences, use “Guest 1”, “the principal”, or initials only.
What each platform does with your data:
- ChatGPT (OpenAI): Free-tier conversations may be used to train future models unless you opt out in Settings → Data Controls → “Improve the model for everyone.” The paid Team and Enterprise plans do not train on your data. If you’re on the free or Plus plan, assume your input is not private.
- Claude (Anthropic): Free and Pro conversations are not used for training by default. Anthropic’s policy is clearer on this than most. Still, don’t paste anything you wouldn’t want stored on a server.
- Gemini (Google): Free-tier conversations may be reviewed by humans and used for training. Gemini Advanced with a Google Workspace plan has stronger privacy protections. Check your Google AI settings.
- Copilot (Microsoft): Enterprise versions have commercial data protection. The free consumer version has weaker guarantees.
For menu planning specifically — if you’re only sharing cuisine themes, ingredients you have, and service style, there’s no privacy risk. That’s generic information. The risk starts when you add guest names, yacht names, itineraries, or personal health details.
Best practice: Use AI freely for menu creation, recipe ideas, provisioning maths, and cultural research — none of that is sensitive. For guest preference management, either anonymise the data first or use a paid plan with explicit no-training guarantees. When in doubt, strip out names and identifying details before pasting.
Want More? The Power Tool
Everything above works in a browser. Copy the prompt, paste it, talk to it. But if you want AI that reads your actual files — guest preference sheets, provisioning lists, past menus — and generates print-ready PDF menu cards, builds handover documents, and turns WhatsApp feedback into guest intelligence, there’s a next level.
Claude Code in the Galley: The Power Tool covers installation, LaTeX PDF generation, WhatsApp intelligence extraction, automated handover notes, and rotational chef sync. Everything a sole chef on a 50-metre yacht needs to run the galley like an operation, not a guessing game.
The short version: The browser gives you menus. Claude Code gives you a system. Read the next article when you’re ready to build one.
The AI is not replacing you. Nothing replaces the chef who knows that this guest prefers their steak slightly more done than they admit, or that the kids will only eat the pasta if it looks exactly like last time. That knowledge lives in you. The AI just handles the parts that don’t need to.
Resources
Claude Code documentation: docs.anthropic.com/claude-code | Claude AI: claude.ai | ChatGPT: chatgpt.com | Gemini: gemini.google.com
Would you use AI in your galley?
Join the conversation